Related Experiment Videos
New method for parallel computation of Hessian matrix of conformational energy function in internal coordinates
Shugo Nakamura1, Daisuke Kyono, Mitsunori Ikeguchi
1Department of Biotechnology, The University of Tokyo, Yayoi 1-1-1, Bunkyo-ku, Tokyo 113-8657, Japan. shugo@bi.a.u-tokyo.ac.jp
Journal of Computational Chemistry
|March 23, 2002
Summary
A new genetic algorithm optimizes parallel Hessian calculations for biomolecules, improving computational efficiency. This method enhances speedup for complex molecular simulations, like Gln-tRNA, overcoming previous scalability limitations.
Area of Science:
- Computational Chemistry
- Biophysics
- Bioinformatics
Background:
- Calculating the Hessian (second derivatives) of conformational energy is crucial for biomolecular simulations.
- Previous parallel algorithms faced scalability issues with increasing processor numbers.
- Efficient Hessian calculation is vital for understanding molecular dynamics and stability.
Purpose of the Study:
- To develop a novel algorithm for parallel Hessian calculation in biomolecules.
- To optimize processor assignment for load balancing and reduced communication costs.
- To improve the scalability of parallel Hessian computation for large biomolecules.
Main Methods:
- A new algorithm divides Hessian matrix calculation into subtasks.
- A genetic algorithm optimizes processor assignment, considering subtask dependencies.
- The method was applied to a glutaminyl transfer RNA (Gln-tRNA) molecule.
Main Results:
- Achieved a speedup of 32.6 times with 60 processors for Gln-tRNA.
- Demonstrated significantly improved scalability compared to previous parallel algorithms.
- Analyzed elapsed times for subtask computation, data sending, and receiving.
Conclusions:
- The proposed genetic algorithm-based approach effectively optimizes parallel Hessian calculations.
- This method overcomes scalability limitations in previous parallel algorithms for biomolecular simulations.
- The optimized parallel computation offers substantial speedup for complex molecular systems.